Laser Welding for Server Liquid Cooling Plates

A liquid cooling plate for a server has to do three things at once: conduct heat from a cold plate or a die into the coolant, contain that coolant at pressure for the life of the machine, and do it with an internal geometry fine enough to be effective. The joints between the plate and the cover, or between the channels and the manifolds, are where all three requirements meet.

Laser welding has become the standard method for making those joints. It produces a narrow, deep weld with a low heat input, which limits distortion in a part whose flatness matters, and it can be automated for a repeatable seam in production. The process is also sensitive to fit up and to surface condition, which is why the design of the joint is as important as the welding parameters.

Why Laser Welding Suits these Plates

The alternative processes each have a drawback. Furnace brazing heats the whole part, which distorts a large thin plate and requires a filler that may not be compatible with the coolant chemistry. Electron beam welding needs a vacuum, which limits the part size. Adhesive bonding cannot meet the pressure and temperature requirements of a server loop. Laser welding concentrates the energy at the joint, so the bulk of the plate stays cool and the flatness is preserved.

Depth of penetration is controlled by the power density and the travel speed, and it can be set to join a thin cover to a thicker base without melting through the cover. That control is what allows a channel wall a few hundred micrometres thick to be welded to a plate several millimetres thick in a single pass, which is not achievable with a process that heats the whole part.

Joint Design and Fit Up

The weld quality depends on the gap between the two parts. A gap that is too large lets the beam pass through the joint without melting both surfaces, producing a weld that looks continuous but has no penetration on one side. A gap that is too small is less of a problem for the weld but harder to achieve consistently across a large part, and the pressing force needed to close it may distort the plate.

A lap joint, in which the cover overlaps a ledge on the base, is the common configuration because it is tolerant of a small mismatch and because it presents a shoulder that the beam can melt into. A butt joint needs much better fit up and is usually reserved for a machined part where the two halves can be held accurately. In both cases the joint should be designed so that the weld bead sits outside the coolant channel, so that a slight variation in penetration does not break into the flow path.

Parameters and Their Control

The parameters are laser power, travel speed, spot size, focus position and shielding gas. Power and speed together set the energy per unit length, and the spot size sets the power density, which determines whether the process operates in conduction mode with a shallow wide weld or in keyhole mode with a deep narrow one. For a cooling plate the keyhole mode is usually wanted, because it produces the penetration needed with a narrow heat affected zone.

The focus position relative to the surface changes the spot size and therefore the power density, so it has to be set and verified for each part. Shielding gas protects the molten pool from oxidation and suppresses the plasma plume that can block the beam. Where the weld is made from one side only, the gas also has to prevent the underside from oxidizing, which may require a backing gas or a controlled atmosphere around the part.

Materials and Their Weldability

Aluminium alloys are the most common material for a cooling plate because of their conductivity and weight. Their weldability varies: a wrought alloy with a low alloying content welds well, while a casting with a high silicon content can be porous and prone to cracking. Where the plate is machined from a wrought alloy and the cover is a casting, the difference in composition and in thermal conductivity has to be allowed for in the parameters.

Copper plates offer better thermal performance and are used where the heat flux is very high, but copper reflects the laser wavelength used by most fibre sources and conducts heat away rapidly, so the process is less efficient. Where copper is used, a shorter wavelength source or a higher power density is needed, and the parameter window is narrower. The choice of material therefore affects the welding process as well as the thermal design, and it should be made with both in mind.

Leak Testing and Quality Assurance

Every welded assembly is leak tested, normally by pressurising the internal volume with a tracer gas and measuring the leakage with a mass spectrometer, or by a pressure decay method for coarser leaks. The test is applied after welding and again after any subsequent machining or finishing operation that could open a weld. A part that passes immediately after welding and fails later usually has a weld with insufficient penetration or with porosity along the seam.

Process monitoring is the other half of quality assurance. Recording the power, the speed and the reflected light or the thermal signature of the weld pool for each part gives a trace that can be compared with a reference, and a deviation indicates a change in fit up, in the material or in the focus. That signature is more sensitive than a visual inspection, and it is available on every unit rather than on a sample. Combined with a destructive test on a sectioned sample, it provides the evidence that the seam, and not only the inspection, is under control, in the same spirit as any other controlled manufacturing process.

Additional Considerations for This Build

Practical attention to leak tightness pays for itself here, because it decides whether the finished board behaves as the drawing intended. Where the requirement is not stated on the fabrication drawing or in the assembly notes, the shop has to assume a default, and that default is rarely the value the design was simulated with. Stating leak tightness explicitly, together with the tolerance that applies, removes the assumption and keeps the result predictable from batch to batch.

Deliberate attention to weld penetration pays for itself here, because it decides whether the finished board behaves as the drawing intended. Where the requirement is not stated on the fabrication drawing or in the assembly notes, the shop has to assume a default, and that default is rarely the value the design was simulated with. Stating weld penetration explicitly, together with the tolerance that applies, removes the assumption and keeps the result predictable from batch to batch.

Process Control and Verification

On a design of this kind, liquid cooling plate is the item that decides how the rest of the board is arranged. The process window is set by the narrowest step in the flow, so an improvement anywhere else shows up as margin rather than as yield until that step is addressed. A short note on the drawing about handling, storage or packaging is often worth more than an extra decimal place on a tolerance.

Related reading: our fabrication notes, board quality and design release notes cover the same ground.

FAQ

Can a leaking weld be repaired? It can be rewelded locally, but the repair introduces a second thermal cycle and a risk of distortion. Where the leak is at the end of a long seam, cutting back and rewelding is more reliable than filling the defect.

Why is fit up so critical? Because the beam has to melt both surfaces at the interface. A gap larger than the beam can bridge produces a weld that joins only the upper part, leaving a leak path underneath.

How is the weld inspected non destructively? By leak testing and by process monitoring of the weld signature. Radiography is possible but is difficult on a thin lap joint because the weld is at an angle to the beam.